Did your Lovable projects
expose credentials or sensitive data?
Lovable accidentally re-enabled access to chat histories on public projects after a backend permissions change. Because developers paste API keys, database URLs, and credentials into AI prompts, public chat histories are higher-risk than public code. Answer 4 questions to assess your exposure.
This is a local self-assessment — no data is sent to our servers. Answers stay in your browser.
Self-audit checklist
Answer 4 questions to assess your exposure and get specific action items.
Background
Lovable's public/private toggle controlled the entire project - chat history, code, and build artifacts. Many users assumed "public" only meant their published app was visible, not the prompts they used to build it.
Vibe-coding prompts typically include API keys pasted for context, database URLs shared to explain errors, service credentials dropped in mid-session, and internal system details. This makes chat histories higher-risk than the generated code.
A February 2026 backend change accidentally re-enabled access to public project chats. Two HackerOne reports were closed without escalation - the triage team read it as intended behavior based on old documentation.
Free tier users before May 2025 could not make projects private. Anyone on any tier who had projects set to public and pasted credentials into chats should assume those chats were accessible during the window.
Based on the Lovable public statement and our full incident analysis.